Research objective
We aimed to model transformation rates simultaneously over attained age or time since diagnosis for MPN patients overall, and by subtypes. Further subanalysis was carried out specific to cytoreductive treatments. Additional summary was also provided for genetic information.
Settings
Study design: cohort study
Data sources:
- the National Cancer Register (for MPN, other hematological malignancies)
- the National Inpatient Register (hospital admissions), the National Outpatient Register (specialist visits) (for MPN and AML/MDS diagnoses)
- the National Causes of Deaths Register (for AML/MDS diagnoses and other causes of deaths)
- the National Prescribed Drugs Register (for cytoreductive drugs)
- the Total Population Register (for demographic data)
- the MPN Quality Register (for genetic data)
Study period: three study periods are covered in this study:
- for describing transformation rates overall and by subtypes, the diagnosis period is 2001 - 2021
- for describing transformation rates by exposure to cytoreductive drugs, the diagnosis period is 2006 - 2021
- for summarising patient numbers by somatic mutations JAK2, CALR, MPL, the diagnosis period is 2008 - 2021
Index date: at 3 months post MPN date date. Individuals were excluded if prior to the index date, they transformed to AML/MDS, developed other hematological malignancy, or they emigrated, or they died.
Event of interest: three types of events were analysed separately: AML/MDS, AML, MDS. For the event AML, patients were not censored if they were diagnosed with MDS prior to AML, whereas for the MDS, patients were censored if they were diagnosed with AML prior to MDS.
End of follow-up: individuals were followed until the outcome of interest or censoring due to above, or due to emigration, due to death from other causes, or due to the end of follow-up (December 31, 2022); whichever occurred first.
Main model
Fitted the following flexible parametric survival model on the log-hazard scale with two time-scales:
\[
\begin{align}
\log h & = s(t_1; \gamma_1) + s(t_2; \gamma_2)
\end{align}
\]
where \(t_1\) is time since the index date and \(t_2\) is attained age. Functions \(s\) represent restricted cubic splines with \(\gamma\) as corresponding parameters. Optimal number of knots for the spline functions were chosen based on AIC and BIC of the models.
Stata code for the model:
stmt mpn, ///
time1(df(3) logtoff) ///
time2(df(3) logtoff start(start_age))
Click here for stata code for predicting thrombosis rates
/* predict hazard rates for each time-point of time1, time2 and age0 */
local n=1
local datalist ""
forvalues a0= 30 30.25 : 90 {
qui clear
qui set obs 401
qui gen age_start =`a0' // a0: age at index
qui range time1 0 20 401 // t1: time on study
qui gen time2 = time1 + age_start // t2: attained age
qui gen _d = .
qui gen _t = .
/* predict hazard rates */
qui predict h, h time1var(time1) time2var(time2) per(1000) ci
qui tempfile temppred`n'
qui save `temppred`n''
local datalist `datalist' `temppred`n''
local n=`n'+1
}
qui clear
qui set obs 0
qui append using `datalist'
qui order age_start time2
keep time1 time2 age_start h*
Results
Cohort characteristics
Table 4.1: Characteristics of Swedish patients diagnosed during 2001-2021 with myeloproliferative neoplasms. MPN=myeloproliferative neoplasm, PV=polycythemia vera, ET=essential thrombocythemia, PMF=primary myelofibrosis, MPN-U=myeloproliferative neoplasm unclassifiable.
| PV (N=7156) | ET (N=6810) | PMF (N=1080) | MPN-U (N=2784) | Total (N=17830) |
|---|
Sex |
|
|
|
|
|
Female | 3178 (44.4%) | 4200 (61.7%) | 452 (41.9%) | 1487 (53.4%) | 9317 (52.3%) |
Male | 3978 (55.6%) | 2610 (38.3%) | 628 (58.1%) | 1297 (46.6%) | 8513 (47.7%) |
Age at diagnosis |
|
|
|
|
|
Median (Q1, Q3) | 70.9 (60.8, 78.2) | 68.2 (55.5, 77.3) | 71.6 (62.7, 78.7) | 71.7 (60.8, 79.3) | 70.1 (59.0, 78.0) |
18-49 | 684 (9.6%) | 1192 (17.5%) | 96 (8.9%) | 317 (11.4%) | 2289 (12.8%) |
50-59 | 1007 (14.1%) | 1011 (14.8%) | 114 (10.6%) | 348 (12.5%) | 2480 (13.9%) |
60-69 | 1718 (24.0%) | 1504 (22.1%) | 280 (25.9%) | 596 (21.4%) | 4098 (23.0%) |
70-79 | 2327 (32.5%) | 1884 (27.7%) | 370 (34.3%) | 885 (31.8%) | 5466 (30.7%) |
80-90 | 1420 (19.8%) | 1219 (17.9%) | 220 (20.4%) | 638 (22.9%) | 3497 (19.6%) |
Calendar period of diagnosis |
|
|
|
|
|
2001-2010 | 3171 (44.3%) | 2635 (38.7%) | 325 (30.1%) | 1675 (60.2%) | 7806 (43.8%) |
2011-2021 | 3985 (55.7%) | 4175 (61.3%) | 755 (69.9%) | 1109 (39.8%) | 10024 (56.2%) |
Event of interest |
|
|
|
|
|
AML/MDS | 291 (4.1%) | 315 (4.6%) | 207 (19.2%) | 407 (14.6%) | 1220 (6.8%) |
AML | 190 (2.7%) | 191 (2.8%) | 135 (12.5%) | 183 (6.6%) | 699 (3.9%) |
MDS | 115 (1.6%) | 166 (2.4%) | 83 (7.7%) | 254 (9.1%) | 618 (3.5%) |
Somatic mutations
Information on gene mutations was obtained from the Swedish MPN Quality Register, which was established in 2008. Due to sparsity and potential unreliability of data, it was not possible to fit any models. Here we provide only summary statistics for JAK2, CALR and MPL mutations, and overall.
Table 6.1: Characteristics of patients with myeloproliferative neoplasms by somatic mutation status (cohort 2008-2021) with event of interest as AML/MDS. MPN=myeloproliferative neoplasm, PV=polycythemia vera, ET=essential thrombocythemia, PMF=primary myelofibrosis, MPN-U=myeloproliferative neoplasm unclassifiable.
| JAK2 | CALR | MPL | JAK2, CALR, MPL |
|
|---|
| Pos (N=4892) | Neg (N=871) | Missing (N=877) | Pos (N=370) | Neg/Missing (N=6270) | Pos (N=107) | Neg/Missing (N=6533) | Triple positive (N=7) | Overall (N=6640) |
|---|
Sex |
|
|
|
|
|
|
|
|
|
Female | 2635 (53.9%) | 450 (51.7%) | 444 (50.6%) | 174 (47.0%) | 3355 (53.5%) | 65 (60.7%) | 3464 (53.0%) | 4 (57.1%) | 3529 (53.1%) |
Male | 2257 (46.1%) | 421 (48.3%) | 433 (49.4%) | 196 (53.0%) | 2915 (46.5%) | 42 (39.3%) | 3069 (47.0%) | 3 (42.9%) | 3111 (46.9%) |
Age at diagnosis |
|
|
|
|
|
|
|
|
|
Median (Q1, Q3) | 70.2 (60.4, 77.1) | 68.6 (56.0, 76.3) | 69.7 (58.1, 78.4) | 66.3 (54.5, 75.7) | 70.1 (59.8, 77.3) | 70.1 (56.4, 76.2) | 69.9 (59.5, 77.2) | 66.4 (58.6, 70.4) | 69.9 (59.5, 77.2) |
18-49 | 560 (11.4%) | 146 (16.8%) | 138 (15.7%) | 65 (17.6%) | 779 (12.4%) | 12 (11.2%) | 832 (12.7%) | 0 (0.0%) | 844 (12.7%) |
50-59 | 636 (13.0%) | 132 (15.2%) | 115 (13.1%) | 70 (18.9%) | 813 (13.0%) | 18 (16.8%) | 865 (13.2%) | 2 (28.6%) | 883 (13.3%) |
60-69 | 1214 (24.8%) | 193 (22.2%) | 193 (22.0%) | 79 (21.4%) | 1521 (24.3%) | 23 (21.5%) | 1577 (24.1%) | 3 (42.9%) | 1600 (24.1%) |
70-79 | 1656 (33.9%) | 262 (30.1%) | 254 (29.0%) | 109 (29.5%) | 2063 (32.9%) | 40 (37.4%) | 2132 (32.6%) | 2 (28.6%) | 2172 (32.7%) |
80-90 | 826 (16.9%) | 138 (15.8%) | 177 (20.2%) | 47 (12.7%) | 1094 (17.4%) | 14 (13.1%) | 1127 (17.3%) | 0 (0.0%) | 1141 (17.2%) |
MPN subtype |
|
|
|
|
|
|
|
|
|
PV | 2213 (45.2%) | 32 (3.7%) | 80 (9.1%) | 4 (1.1%) | 2321 (37.0%) | 1 (0.9%) | 2324 (35.6%) | 0 (0.0%) | 2325 (35.0%) |
ET | 1823 (37.3%) | 546 (62.7%) | 502 (57.2%) | 249 (67.3%) | 2622 (41.8%) | 71 (66.4%) | 2800 (42.9%) | 4 (57.1%) | 2871 (43.2%) |
PMF | 440 (9.0%) | 200 (23.0%) | 170 (19.4%) | 96 (25.9%) | 714 (11.4%) | 25 (23.4%) | 785 (12.0%) | 0 (0.0%) | 810 (12.2%) |
MPN-U | 416 (8.5%) | 93 (10.7%) | 125 (14.3%) | 21 (5.7%) | 613 (9.8%) | 10 (9.3%) | 624 (9.6%) | 3 (42.9%) | 634 (9.5%) |
Calendar period of diagnosis |
|
|
|
|
|
|
|
|
|
2008-2014 | 2080 (42.5%) | 145 (16.6%) | 696 (79.4%) | 16 (4.3%) | 2905 (46.3%) | 8 (7.5%) | 2913 (44.6%) | 0 (0.0%) | 2921 (44.0%) |
2015-2021 | 2812 (57.5%) | 726 (83.4%) | 181 (20.6%) | 354 (95.7%) | 3365 (53.7%) | 99 (92.5%) | 3620 (55.4%) | 7 (100.0%) | 3719 (56.0%) |
Event of interest |
|
|
|
|
|
|
|
|
|
AML/MDS | 308 (6.3%) | 50 (5.7%) | 102 (11.6%) | 7 (1.9%) | 453 (7.2%) | 5 (4.7%) | 455 (7.0%) | 1 (14.3%) | 460 (6.9%) |
AML | 193 (3.9%) | 34 (3.9%) | 63 (7.2%) | 5 (1.4%) | 285 (4.5%) | 2 (1.9%) | 288 (4.4%) | 1 (14.3%) | 290 (4.4%) |
MDS | 137 (2.8%) | 19 (2.2%) | 44 (5.0%) | 2 (0.5%) | 198 (3.2%) | 3 (2.8%) | 197 (3.0%) | 0 (0.0%) | 200 (3.0%) |
Statistical Software
The statistical software used for the analysis was R version 4.3.1.